Experimental lora using a new method of object training. Same intent and use as current Penis_xl lora still indented for stable diffusion and comfy.

I used 16 *pov from the top (photos of my erect penis) Including: a range of lighting, hardness, veins, blurry (for more real photo effect), and mostly retracted foreskin. All cropped out from original images to only have a pure white background. Not sure if this was the best method but so far I am getting results that I like.

The idea was to train it as an object so it could be used more in POV scenarios (dick picks, cock worships, dildos, random objects, monster cocks, funny cock related content, etc) I am defiantly not nearly as skilled in Lora training as this community, but I am trying to learn.

Would love to see your creations, please share them :)

I have noticed that this model is less dominant than my current lora however the results do seem promising.

Use notes:

Try using only one of the reference words such as "penis" or " penisxl"

I think it works best so far between 0.7 - 0.86

Try something like this?

POV, photo, nsfw, sexy adult bengali model naked (holding a 9_inch_penis), <lora:PenisXL:0.86>, cumming, uncensored, erotic, sad expression, (eyes closed), glamour portrait, bedroom, neon lighting, glossy, wet, (runny_mascara), (smudged_makeup), dark background with moody lighting neon, blissful, (wearing glasses), high ponytail hair, sweaty, medium_breasts, dripping cum

OR THIS?

POV, photo, nsfw, handsome male model ((holding a 9_inch_penis)), <lora:PenisXL:0.82>, cumming, uncensored, erotic, blissful expression, (eyes closed), glamour portrait, bedroom, neon lighting, glossy, wet, dark background with moody lighting neon, muscular, short beard, buzzed hair, sweaty, dripping cum

Model: Photo XL

Steps: 35

Sampler: DPM++ 2S a Karras

CFG scale: 7.5 - 8.5

Denoising strength: 0.33

Width: 768 x Height: 1160

I really like this model for realistic photos:

Photo XL - PhotoXLV1.2 | Stable Diffusion Checkpoint | Civitai

For those who are curious about the accuracy and reference images used, don't be shy just ask and I'll show you.